q-viper

    
def preprocess(bgr_img):#gray image   
    img = bgr_img[:]
    blur = cv2.GaussianBlur(img,(5,5),0)
    ret,th_img = cv2.threshold(blur,0,255,cv2.THRESH_BINARY_INV+cv2.THRESH_OTSU) #converts black to white and inverse

    rows, cols = th_img.shape
    bg_test = np.array([th_img[i][i] for i in range(5)])
    if bg_test.all() == 0:
        text_color = 255

q-viper

def borders(here_img, thresh):
    size = here_img.shape
    check = int(115 * size[0] / 600)
    image = here_img[:]
    top, bottom = 0, size[0] - 1
    #plt.imshow(image)
    #plt.show()
    shape = size

q-viper

def segmentation(bordered, thresh):
    try:
        shape = bordered.shape
        check = int(50 * shape[0] / 320)
        image = bordered[:]
        image = image[check:].T
        shape = image.shape
        #plt.imshow(image)
        #plt.show()

q-viper

def localize(main_image, gray_img, localized, bc, show):
    #open the template as gray scale image
        template = localized
        #print(template.shape)
        width, height = template.shape[::-1] #get the width and height
        #match the template using cv2.matchTemplate
        match = cv2.matchTemplate(gray_img, template, cv2.TM_CCOEFF_NORMED)
        threshold = 0.8
        position = np.where(match >= threshold) #get the location of template in the image
        for point in zip(*position[::-1]): #draw the rectangle around the matched template

q-viper

def detect_text(main_image, gray_img, localized, bc):        
        cimg = cv2.resize(localized, (30, 30))
        bordersize = 1
        nimg = cv2.copyMakeBorder(cimg, top=bordersize, bottom=bordersize, left=bordersize, right=bordersize, borderType=cv2.BORDER_CONSTANT, value=[255-bc, 0, 0])

        return main_image, nimg

q-viper

from preprocess import preprocess, detect_text, localize
from predictor import prediction
import numpy as np
import matplotlib.pyplot as plt
import cv2

def recognition(gray_image, show):
    segments, template, th_img, text_color = preprocess(gray_image)
    labels = []
    accuracy = []

q-viper

import numpy as np
from keras.models import model_from_json
from keras.models import load_model

def prediction(img):
    # load json and create model
    json_file = open('cnn2\cnn2.json', 'r')
    
    loaded_model_json = json_file.read()
    json_file.close()

q-viper

import cv2
from recognition import recognition
import numpy as np
import time
import matplotlib.pyplot as plt

def camera(flag):
   # choice = print("Click spacebar for photo and anything else for video.\n")
    orig = 1
    cap = cv2.VideoCapture(0)

q-viper

from recognition import recognition
import cv2
import matplotlib.pyplot as plt
from video_test import camera
import time

try: 
    test = input('Please enter the image directory with name.\n')
    test = cv2.imread(test, 0)
    plt.imshow(cv2.cvtColor(test, cv2.COLOR_GRAY2RGB))

q-viper

def show(img, figsize=(10, 10)):
    figure = plt.figure(figsize=figsize)
    plt.imshow(img)
    plt.xticks([])
    plt.yticks([])
    plt.show()


cap = cv2.VideoCapture(0) #run camera with device 
imgs = [] #list to hold frames